No OneTemporary
Actions

Size

45 KB

Referenced Files

None

Subscribers

None

View Options

	diff --git a/.formatter.exs b/.formatter.exs
	index d2cda26..2acd350 100644
	--- a/.formatter.exs
	+++ b/.formatter.exs
	@@ -1,4 +1,7 @@
	+# SPDX-FileCopyrightText: 2019-2022 Pleroma Authors <https://pleroma.social>
	+# SPDX-License-Identifier: LGPL-2.1-only
	+
	# Used by "mix format"
	[
	inputs: ["{mix,.formatter}.exs", "{config,lib,test}/*/.{ex,exs}"]
	]
	diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml
	index ee1665a..efe5acd 100644
	--- a/.gitlab-ci.yml
	+++ b/.gitlab-ci.yml
	@@ -1,38 +1,41 @@
	+# SPDX-FileCopyrightText: 2019-2022 Pleroma Authors <https://pleroma.social>
	+# SPDX-License-Identifier: LGPL-2.1-only
	+
	image: elixir:1.12-alpine

	variables:
	MIX_ENV: test
	GIT_SUBMODULE_STRATEGY: recursive

	cache:
	key: ${CI_COMMIT_REF_SLUG}
	paths:
	- deps
	- _build

	stages:
	- test
	- publish

	before_script:
	- apk add build-base cmake
	- mix local.hex --force
	- mix local.rebar --force
	- mix deps.get --only test
	- mix compile --force

	lint:
	stage: test
	script:
	- mix format --check-formatted

	unit-testing:
	stage: test
	coverage: '/(\d+\.\d+\%) \\| Total/'
	script:
	- mix test --trace --preload-modules --cover

	dialyzer:
	stage: test
	script:
	- mix dialyzer
	diff --git a/.gitmodules b/.gitmodules
	index c842c8c..2dea083 100644
	--- a/.gitmodules
	+++ b/.gitmodules
	@@ -1,7 +1,8 @@
	# SPDX-FileCopyrightText: 2017-2019 myhtmlex authors <https://github.com/Overbryd/myhtmlex>
	+# SPDX-FileCopyrightText: 2019-2022 Pleroma Authors <https://pleroma.social>
	# SPDX-License-Identifier: LGPL-2.1-only

	[submodule "c_src/lexbor"]
	path = c_src/lexbor
	url = https://github.com/lanodan/lexbor.git
	branch = bugfix/gcc-10
	diff --git a/CHANGELOG.md b/CHANGELOG.md
	index e634b71..8a651eb 100644
	--- a/CHANGELOG.md
	+++ b/CHANGELOG.md
	@@ -1,58 +1,63 @@
	+<!--
	+SPDX-FileCopyrightText: 2019-2022 Pleroma Authors <https://pleroma.social>
	+SPDX-License-Identifier: LGPL-2.1-only
	+-->
	+
	# Changelog
	All notable changes to this project will be documented in this file.

	The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/).

	## [2.0.4] - 2020-09-01
	#### Fixed
	- Regression of the gcc 10 fix in the hex package, due to an outdated submodule on the publishing machine

	## [2.0.3] - 2020-08-30
	### Changed
	- Improved error message on make errors

	## [2.0.2] - 2020-08-26
	### Fixed
	- lexbor compilation errors with gcc 10
	- Inability to use the library with distillery releases due to priv dir being in the source code

	## [2.0.1] - 2020-08-04
	### Fixed
	- Build failures using the hex package due to CMake cache files accidentally included there

	### Added
	- Changelog is now available at hexdocs


	## [2.0.0] - 2020-08-01
	### Changed
	- Breaking: CMake is now required at compile-time due to it being lexbor's build system
	- Breaking: namespaces are no longer automatically appended. i.e `<svg> </svg>` will be `{"svg", [], []}` instead of `{"svg:svg", [], []}`
	- Breaking: when using `:nil_self_closing` flag, only valid [void elements](https://html.spec.whatwg.org/#void-elements) will have `nil` in children
	- Now deprecated myhtml was switched to [lexbor](https://github.com/lexbor/lexbor)
	- The worker process now communicates with the node via stdio, instead of TCP, which was known to cause issues
	on BSD systems

	### Added
	- `FastHtml.Pool` for fast_html workers. There is a default pool of `System.schedulers_online/0` workers, but a custom pool can be started if desired, or it can be disabled altogether. See `FastHtml.Pool` module documentation for more info

	## [1.0.3] - 2020-02-10
	### Fixed
	- C-Node not respawning after being killed.

	## [1.0.2] - 2020-02-10
	### Fixed
	- Incorrect behavior when parsing empty attribute values. Instead of an empty string the attribute name was returned.

	## [1.0.1] - 2019-12-11
	### Added
	- `:fast_html.decode_fragment`
	### Fixed
	- Errors from C-Node not being reported, timing out instead

	## [1.0.0] - 2019-12-02
	### Changed
	- BREAKING: `:fast_html.decode` now returns an array of nodes at the top level, instead of a single node. This was done because it's possible to have more than one root node, for example in (`<!-- a comment --> <html> </html>` both the comment and the `html` tag are root nodes).

	### Fixed
	- Worker going into infinite loop when decoding a document with more than one root node.
	diff --git a/Makefile b/Makefile
	index 87c0f41..5851262 100644
	--- a/Makefile
	+++ b/Makefile
	@@ -1,59 +1,60 @@
	# SPDX-FileCopyrightText: 2017-2019 myhtmlex authors <https://github.com/Overbryd/myhtmlex>
	+# SPDX-FileCopyrightText: 2019-2022 Pleroma Authors <https://pleroma.social>
	# SPDX-License-Identifier: LGPL-2.1-only

	MIX = mix
	CMAKE = cmake
	CNODE_CFLAGS = -g -O2 -std=c99 -pedantic -Wcomment -Wextra -Wno-old-style-declaration -Wall

	# ignore unused parameter warnings
	CNODE_CFLAGS += -Wno-unused-parameter

	# set erlang include path
	ERLANG_PATH = $(shell erl -eval 'io:format("~s", [lists:concat([code:root_dir(), "/erts-", erlang:system_info(version)])])' -s init stop -noshell)
	CNODE_CFLAGS += -I$(ERLANG_PATH)/include

	# expecting myhtml as a submodule in c_src/
	# that way we can pin a version and package the whole thing in hex
	# hex does not allow for non-app related dependencies.
	LXB_PATH = c_src/lexbor
	LXB_STATIC = $(LXB_PATH)/liblexbor_static.a
	CNODE_CFLAGS += -I$(LXB_PATH)/source
	# avoid undefined reference errors to phtread_mutex_trylock
	CNODE_CFLAGS += -lpthread

	# C-Node
	ERL_INTERFACE = $(wildcard $(ERLANG_PATH)/../lib/erl_interface-*)
	CNODE_CFLAGS += -L$(ERL_INTERFACE)/lib
	CNODE_CFLAGS += -I$(ERL_INTERFACE)/include

	CNODE_LDFLAGS =

	ifeq ($(OTP22_DEF),YES)
	CNODE_CFLAGS += -DOTP_22_OR_NEWER
	else
	CNODE_LDFLAGS += -lerl_interface
	endif

	CNODE_LDFLAGS += -lei -pthread

	.PHONY: all

	all: priv/fasthtml_worker

	$(LXB_STATIC): $(LXB_PATH)
	# Sadly, build components separately seems to sporadically fail
	cd $(LXB_PATH); cmake -DLEXBOR_BUILD_SEPARATELY=OFF -DLEXBOR_BUILD_SHARED=OFF
	$(MAKE) -C $(LXB_PATH)

	priv/fasthtml_worker: c_src/fasthtml_worker.c $(LXB_STATIC)
	mkdir -p priv
	$(CC) -o $@ $< $(LXB_STATIC) $(CNODE_CFLAGS) $(CNODE_LDFLAGS)

	clean: clean-myhtml
	$(RM) -r priv/myhtmlex*
	$(RM) priv/fasthtml_worker
	$(RM) myhtmlex-*.tar
	$(RM) -r package-test

	clean-myhtml:
	$(MAKE) -C $(MYHTML_PATH) clean
	diff --git a/README.md b/README.md
	index c5d5f9d..d01eb3b 100644
	--- a/README.md
	+++ b/README.md
	@@ -1,34 +1,35 @@
	<!--
	SPDX-FileCopyrightText: 2017-2019 myhtmlex authors <https://github.com/Overbryd/myhtmlex>
	+SPDX-FileCopyrightText: 2019-2022 Pleroma Authors <https://pleroma.social>
	SPDX-License-Identifier: LGPL-2.1-only
	-->

	# FastHTML

	A C Node wrapping lexborisov's [myhtml](https://github.com/lexborisov/myhtml).
	Primarily used with [FastSanitize](https://git.pleroma.social/pleroma/fast_sanitize).

	* Available as a hex package: `{:fast_html, "~> 2.0"}`
	* [Documentation](https://hexdocs.pm/fast_html/fast_html.html)

	## Benchmarks

	The following table provides median times it takes to decode a string to a tree for html parsers that can be used from Elixir. Benchmarks were conducted on a machine with an `AMD Ryzen 9 3950X (32) @ 3.500GHz` CPU and 32GB of RAM. The `mix fast_html.bench` task can be used for running the benchmark by yourself.

	\| File/Parser \| fast_html (Port) \| mochiweb_html (erlang) \| html5ever (Rust NIF) \| Myhtmlex (NIF)¹ \|
	\|----------------------\|--------------------\|------------------------\|----------------------\|----------------\|
	\| document-large.html (6.9M) \| 125.12 ms \| 1778.34 ms \| 395.21 ms \| 327.17 ms \|
	\| document-medium.html (85K) \| 1.93 ms \| 12.10 ms \| 4.74 ms \| 3.82 ms \|
	\| document-small.html (25K)\| 0.50 ms \| 2.76 ms \| 1.72 ms \| 1.19 ms \|
	\| fragment-large.html (33K)\| 0.93 ms \| 4.78 ms \| 2.34 ms \| 2.15 ms \|
	\| fragment-small.html² (757B)\| 44.60 μs \| 42.13 μs \| 43.58 μs \| 289.71 μs \|

	Full benchmark output can be seen in [this snippet](https://git.pleroma.social/pleroma/elixir-libraries/fast_html/snippets/3128)

	1. Myhtmlex has a C-Node mode, but it wasn't benchmarked here because it segfaults on `document-large.html`
	2. The slowdown on `fragment-small.html` is due to Port overhead. Unlike html5ever and Myhtmlex in NIF mode, `fast_html` has the parser process isolated and communicates with it over stdio, so even if a fatal crash in the parser happens, it won't bring down the entire VM.

	## Contribution / Bug Reports

	* Please make sure you do `git submodule update` after a checkout/pull
	* The project aims to be fully tested
	diff --git a/c_src/fasthtml_worker.c b/c_src/fasthtml_worker.c
	index 566672e..0030b2b 100644
	--- a/c_src/fasthtml_worker.c
	+++ b/c_src/fasthtml_worker.c
	@@ -1,534 +1,537 @@
	+// SPDX-FileCopyrightText: 2019-2022 Pleroma Authors <https://pleroma.social>
	+// SPDX-License-Identifier: LGPL-2.1-only
	+
	#include <stdlib.h>
	#include <stdbool.h>
	#include <stdio.h>
	#include <string.h>
	#include <stdarg.h>
	#include <unistd.h>
	#include <sys/types.h>
	#include <sys/socket.h>
	#include <netinet/in.h>
	#include <arpa/inet.h>
	#include <errno.h>
	#include <ctype.h>

	#ifndef _REENTRANT
	#define _REENTRANT /* For some reason __erl_errno is undefined unless _REENTRANT is defined */
	#endif

	#include "ei.h"
	#ifndef OTP_22_OR_NEWER
	# include "erl_interface.h"
	#endif

	#define HEADER_SIZE 4

	#include <lexbor/html/html.h>
	#include "tstack.h"

	#ifdef __GNUC__
	# define AFP(x, y) __attribute__((format (printf, x, y)))
	#else
	# define AFP(x, y)
	#endif

	#ifdef __GNUC__
	# define NORETURN __attribute__((noreturn))
	#else
	# define NORETURN
	#endif

	typedef struct _state_t {
	ei_x_buff buffer;
	} state_t;

	typedef enum parse_flags_e {
	FLAG_HTML_ATOMS = 1 << 0,
	FLAG_NIL_SELF_CLOSING = 1 << 1,
	FLAG_COMMENT_TUPLE3 = 1 << 2
	} parse_flags_t;

	char* read_packet(int *len);
	static void handle_send(state_t * state);
	static void err_term(ei_x_buff * response, const char * error_atom);
	static parse_flags_t decode_parse_flags(state_t * state, int arity);
	static void decode(state_t * state, ei_x_buff * response, lxb_html_document_t document, bool fragment, lxb_dom_element_t context_element, lxb_char_t * bin_data, size_t bin_size, parse_flags_t parse_flags);

	static void build_tree(ei_x_buff * response, lxb_dom_node_t* tree, parse_flags_t parse_flags);
	static void prepare_node_attrs(ei_x_buff * response, lxb_dom_node_t* node);

	static inline char * lowercase(char * c);

	static void panic(const char *fmt, ...) AFP(1, 2);
	static void panic(const char *fmt, ...) {
	char buf[4096];
	va_list va;

	va_start (va, fmt);
	vsnprintf (buf, sizeof buf, fmt, va);
	va_end (va);

	fprintf (stderr, "fast_html worker: error: %s\n", buf);
	exit (EXIT_FAILURE);
	}

	int main(int argc, const char *argv[]) {
	state_t* state = calloc (1, sizeof(state_t));

	#ifdef OTP_22_OR_NEWER
	// initialize erlang client library
	ei_init ();
	#else
	erl_init (NULL, -1);
	#endif

	ei_x_new (&state->buffer);

	fflush (stdout);

	while (true) {
	int len;
	char* buf = read_packet(&len);
	ei_x_free(&state->buffer);
	state->buffer.index = 0;
	state->buffer.buff = buf;
	state->buffer.buffsz = len;
	handle_send (state);
	}

	// shutdown: free all state
	ei_x_free (&state->buffer);
	free (state);

	return EXIT_SUCCESS;
	}


	/*
	* Reads a packet from Erlang. The packet must be a standard {packet, 2}
	* packet. This function aborts if any error is detected (including EOF).
	*
	* Returns: The number of bytes in the packet.
	*/

	char read_packet(int len)
	{

	char* io_buf = NULL; /* Buffer for file i/o. */
	unsigned char header[HEADER_SIZE];
	uint32_t packet_length; /* Length of current packet. */
	uint32_t bytes_read;
	uint32_t total_bytes_read;

	/*
	* Read the packet header.
	*/

	total_bytes_read = read(STDIN_FILENO, header, HEADER_SIZE);

	if (total_bytes_read == 0) {
	exit(0);
	}
	if (total_bytes_read != HEADER_SIZE) {
	panic("Failed to read packet header, read: %d\n", total_bytes_read);
	}

	/*
	* Get the length of this packet.
	*/

	packet_length = 0;

	for (int i = 0; i < HEADER_SIZE; i++)
	packet_length = (packet_length << 8) \| header[i];

	*len=packet_length;

	if ((io_buf = (char *) malloc(packet_length)) == NULL) {
	panic("insufficient memory for i/o buffer of size %d\n", packet_length);
	}

	/*
	* Read the packet itself.
	*/

	total_bytes_read = 0;

	while((bytes_read = read(STDIN_FILENO, (io_buf + total_bytes_read), (packet_length - total_bytes_read))))
	total_bytes_read += bytes_read;

	if (total_bytes_read != packet_length) {
	free(io_buf);
	panic("couldn't read packet of length %d, read: %d\r\n",
	packet_length, total_bytes_read);
	}

	return io_buf;
	}

	// handle ERL_SEND message type.
	// we expect a tuple with arity of 3 or 4 in state->buffer.
	// we expect the first argument to be an atom (`decode` or `decode_fragment`),
	// the second argument to be the HTML payload, and the
	// third argument to be the argument list.
	// In case of `decode_fragment`, the fourth argument should be
	// the context tag name.
	// any other message: respond with an {error, unknown_call} tuple.
	static void handle_send (state_t * state)
	{
	// response holds our response, prepare it
	ei_x_buff response;

	ei_x_new (&response);

	// check the protocol version, if it's unsupported, panic
	int version;
	if (ei_decode_version (state->buffer.buff, &state->buffer.index, &version) < 0)
	panic ("malformed message - bad version (%d).", version);

	// decode the tuple header
	int arity;
	if (ei_decode_tuple_header (state->buffer.buff, &state->buffer.index, &arity) < 0)
	{
	err_term (&response, "badmatch");
	goto out;
	}

	char atom[MAXATOMLEN];
	if (ei_decode_atom (state->buffer.buff, &state->buffer.index, atom) < 0)
	{
	err_term (&response, "badmatch");
	goto out;
	}

	bool fragment = false;
	if (strcmp (atom, "decode"))
	{
	if (strcmp (atom, "decode_fragment")) {
	err_term (&response, "unknown_call");
	goto out;
	} else if (arity != 4) {
	err_term (&response, "badmatch");
	goto out;
	} else {
	fragment = true;
	}
	} else if (arity != 3) {
	err_term (&response, "badmatch");
	goto out;
	}

	// the next argument should be a binary, allocate it dynamically.
	int bin_type, bin_size;
	if (ei_get_type (state->buffer.buff, &state->buffer.index, &bin_type, &bin_size) < 0)
	panic ("failed to decode binary size in message");

	// verify the type
	if (bin_type != ERL_BINARY_EXT)
	{
	err_term (&response, "badmatch");
	goto out;
	}

	// decode the binary
	char * bin_data = calloc (1, bin_size + 1);
	if (ei_decode_binary (state->buffer.buff, &state->buffer.index, bin_data, NULL) < 0)
	panic ("failed to decode binary in message");

	// next should be the options list
	if (ei_decode_list_header (state->buffer.buff, &state->buffer.index, &arity) < 0)
	panic ("failed to decode options list header in message");

	parse_flags_t parse_flags = decode_parse_flags (state, arity);

	// Lists with items always have an empty list as their tail
	if (arity != 0)
	if (ei_decode_list_header (state->buffer.buff, &state->buffer.index, &arity) < 0)
	panic ("failed to decode empty list header after option list in message");

	lxb_html_document_t *document = lxb_html_document_create();
	lxb_dom_element_t *context_element = NULL;

	// if we are parsing a fragment, context tag name should come next
	if (fragment) {
	int context_bin_type, context_bin_size;
	if (ei_get_type (state->buffer.buff, &state->buffer.index, &context_bin_type, &context_bin_size) < 0)
	panic ("failed to decode binary size in message");

	// verify the type
	if (context_bin_type != ERL_BINARY_EXT)
	{
	err_term (&response, "badmatch");
	goto out;
	}

	// decode the binary
	char* context_bin_data = calloc (1, context_bin_size + 1);
	if (ei_decode_binary (state->buffer.buff, &state->buffer.index, context_bin_data, NULL) < 0)
	panic ("failed to decode context binary in message");

	context_element = lxb_dom_document_create_element(&document->dom_document, (lxb_char_t*) context_bin_data, context_bin_size, NULL);
	free (context_bin_data);
	}

	if (context_element && lxb_dom_element_tag_id(context_element) >= LXB_TAG__LAST_ENTRY) {
	err_term (&response, "unknown_context_tag");
	} else {
	decode (state, &response, document, fragment, context_element, (lxb_char_t *) bin_data, bin_size, parse_flags);
	}
	lxb_html_document_destroy(document);
	free (bin_data);

	out: ;
	// send response
	unsigned char header[HEADER_SIZE];
	uint32_t size = (uint32_t) response.index;

	for (int i = HEADER_SIZE-1; i != -1; i--) {
	header[i] = (unsigned char) size & 0xFF;
	size = size >> 8;
	}

	write(STDOUT_FILENO, header, sizeof(header));
	write(STDOUT_FILENO, response.buff, response.index);
	// free response
	ei_x_free (&response);
	return;
	}

	static void err_term (ei_x_buff * response, const char * error_atom)
	{
	response->index = 0;
	ei_x_encode_version (response);
	ei_x_encode_tuple_header (response, 2);
	ei_x_encode_atom (response, "error");
	ei_x_encode_atom (response, error_atom);
	}

	static parse_flags_t decode_parse_flags (state_t * state, int arity)
	{
	parse_flags_t parse_flags = 0;

	for (int i = 0; i < arity; i++)
	{
	char atom[MAXATOMLEN];

	if (ei_decode_atom (state->buffer.buff, &state->buffer.index, atom) < 0)
	continue;

	if (! strcmp ("html_atoms", atom))
	parse_flags \|= FLAG_HTML_ATOMS;
	else if (! strcmp ("nil_self_closing", atom))
	parse_flags \|= FLAG_NIL_SELF_CLOSING;
	else if (! strcmp ("comment_tuple3", atom))
	parse_flags \|= FLAG_COMMENT_TUPLE3;
	}

	return parse_flags;
	}

	static void decode(state_t * state, ei_x_buff * response, lxb_html_document_t document, bool fragment, lxb_dom_element_t context_element, lxb_char_t * bin_data, size_t bin_size, parse_flags_t parse_flags)
	{
	// parse tree
	lxb_status_t status;
	lxb_dom_node_t *node;

	if (fragment) {
	node = lxb_html_document_parse_fragment(document, context_element, bin_data, bin_size);
	status = (node == NULL)? LXB_STATUS_ERROR : LXB_STATUS_OK;
	} else {
	status = lxb_html_document_parse(document, bin_data, bin_size);
	node = lxb_dom_interface_node(document);
	}

	if (status != LXB_STATUS_OK)
	{
	err_term (response, "parse_failed");
	return;
	}

	// build tree
	build_tree (response, node, parse_flags);
	}

	// a tag is sent as a tuple:
	// - a string or atom for the tag name
	// - an attribute list
	// - a children list
	// in this function, we prepare the atom and complete attribute list
	static void prepare_tag_header (ei_x_buff * response, const char * tag_string, lxb_dom_node_t* node, parse_flags_t parse_flags)
	{
	lxb_tag_id_t tag_id = lxb_dom_node_tag_id(node);

	ei_x_encode_tuple_header (response, 3);

	if (! (parse_flags & FLAG_HTML_ATOMS) \|\| (tag_id == LXB_TAG__UNDEF \|\| tag_id >= LXB_TAG__LAST_ENTRY))
	ei_x_encode_binary (response, tag_string, strlen (tag_string));
	else
	ei_x_encode_atom (response, tag_string);

	prepare_node_attrs (response, node);
	}

	// prepare an attribute node
	static void prepare_node_attrs(ei_x_buff * response, lxb_dom_node_t* node)
	{
	lxb_dom_attr_t *attr;

	for (attr = lxb_dom_element_first_attribute(lxb_dom_interface_element(node)); attr != NULL; attr = lxb_dom_element_next_attribute(attr))
	{
	size_t attr_name_len;
	char attr_name = (char) lxb_dom_attr_qualified_name(attr, &attr_name_len);
	size_t attr_value_len;
	const char attr_value = (char) lxb_dom_attr_value(attr, &attr_value_len);

	/* guard against poisoned attribute nodes */
	if (! attr_name_len)
	continue;

	ei_x_encode_list_header (response, 1);
	ei_x_encode_tuple_header (response, 2);
	ei_x_encode_binary (response, attr_name, attr_name_len);

	ei_x_encode_binary (response, attr_value, attr_value_len);
	}

	ei_x_encode_empty_list (response);
	}

	// dump a comment node
	static void prepare_comment (ei_x_buff * response, const char * node_comment, size_t comment_len, parse_flags_t parse_flags)
	{
	ei_x_encode_tuple_header (response, parse_flags & FLAG_COMMENT_TUPLE3 ? 3 : 2);
	ei_x_encode_atom (response, "comment");

	if (parse_flags & FLAG_COMMENT_TUPLE3)
	ei_x_encode_list_header (response, 0);

	ei_x_encode_binary (response, node_comment, comment_len);
	}

	#ifdef DEBUG_LIST_MANIP

	#define EMIT_LIST_HDR \
	printf ("list hdr for node %p\n", current_node); \
	fflush (stdout); \
	ei_x_encode_list_header (response, 1)

	#define EMIT_EMPTY_LIST_HDR \
	printf ("list empty for node %p\n", current_node); \
	fflush (stdout); \
	ei_x_encode_list_header (response, 0)

	#define EMIT_LIST_TAIL \
	printf ("list tail for node %p\n", current_node); \
	fflush (stdout); \
	ei_x_encode_empty_list (response)

	#else

	#define EMIT_LIST_HDR ei_x_encode_list_header (response, 1)
	#define EMIT_EMPTY_LIST_HDR ei_x_encode_list_header (response, 0)
	#define EMIT_LIST_TAIL ei_x_encode_empty_list (response)

	#endif

	static void build_tree (ei_x_buff * response, lxb_dom_node_t* node, parse_flags_t parse_flags)
	{
	tstack stack;
	tstack_init (&stack, 30);
	tstack_push (&stack, node);

	lxb_dom_node_t* current_node = node->first_child;

	// ok we're going to send an actual response so start encoding it
	response->index = 0;
	ei_x_encode_version (response);
	ei_x_encode_tuple_header(response, 2);
	ei_x_encode_atom(response, "ok");

	if (current_node == NULL) {
	EMIT_EMPTY_LIST_HDR;
	EMIT_LIST_TAIL;
	}
	while (current_node != NULL)
	{
	if (current_node->type == LXB_DOM_NODE_TYPE_TEXT)
	{
	size_t text_len;
	const char * node_text = (char*) lxb_dom_node_text_content(current_node, &text_len);
	EMIT_LIST_HDR;
	ei_x_encode_binary (response, node_text, text_len);
	}
	else if (current_node->type == LXB_DOM_NODE_TYPE_COMMENT)
	{
	size_t comment_len;
	const char* node_comment = (char*) lxb_dom_node_text_content(current_node, &comment_len);

	EMIT_LIST_HDR;
	prepare_comment(response, node_comment, comment_len, parse_flags);
	}
	else if(current_node->type == LXB_DOM_NODE_TYPE_ELEMENT)
	{
	// get name of tag
	size_t tag_name_len;
	const char tag_name = (char) lxb_dom_element_qualified_name(lxb_dom_interface_element(current_node), &tag_name_len);
	EMIT_LIST_HDR;
	prepare_tag_header (response, tag_name, current_node, parse_flags);

	if (current_node->first_child)
	{
	tstack_push (&stack, current_node);
	current_node = current_node->first_child;

	continue;
	}
	else
	{
	if (parse_flags & FLAG_NIL_SELF_CLOSING && lxb_html_tag_is_void(lxb_dom_node_tag_id(current_node))) {

	#ifdef DEBUG_LIST_MANIP
	printf ("self-closing tag %s emit nil?\n", tag_string); fflush (stdout);
	#endif
	ei_x_encode_atom (response, "nil");
	}
	else
	{
	EMIT_EMPTY_LIST_HDR;
	}
	}
	}

	if (current_node->next)
	current_node = current_node->next;
	else
	{
	while (! current_node->next && stack.used != 0)
	{
	EMIT_LIST_TAIL;
	current_node = tstack_pop (&stack);
	}

	if (current_node->next)
	current_node = current_node->next;
	}

	// are we at root?
	if (current_node == node)
	break;
	}

	tstack_free (&stack);
	}

	static inline char * lowercase(char* c)
	{
	char * p = c;

	while (*p)
	{
	p = tolower ((unsigned char) p);
	p++;
	}

	return c;
	}
	diff --git a/c_src/tstack.h b/c_src/tstack.h
	index ef3f007..1f6223b 100644
	--- a/c_src/tstack.h
	+++ b/c_src/tstack.h
	@@ -1,38 +1,41 @@
	+// SPDX-FileCopyrightText: 2019-2022 Pleroma Authors <https://pleroma.social>
	+// SPDX-License-Identifier: LGPL-2.1-only
	+
	#ifndef TSTACK_H
	#define TSTACK_H

	#define GROW_BY 30

	typedef struct {
	lxb_dom_node_t **data;
	size_t used;
	size_t size;
	} tstack;

	void tstack_init(tstack *stack, size_t initial_size) {
	stack->data = (lxb_dom_node_t *) malloc(initial_size sizeof(lxb_dom_node_t *));
	stack->used = 0;
	stack->size = initial_size;
	}

	void tstack_free(tstack *stack) {
	free(stack->data);
	}

	void tstack_resize(tstack *stack, size_t new_size) {
	stack->data = (lxb_dom_node_t *) realloc(stack->data, new_size sizeof(lxb_dom_node_t *));
	stack->size = new_size;
	}

	void tstack_push(tstack stack, lxb_dom_node_t element) {
	if(stack->used == stack->size) {
	tstack_resize(stack, stack->size + GROW_BY);
	}
	stack->data[stack->used++] = element;
	}

	lxb_dom_node_t * tstack_pop(tstack *stack) {
	return stack->data[--(stack->used)];
	}

	#endif
	diff --git a/lib/fast_html.ex b/lib/fast_html.ex
	index 723778e..0fc92bc 100644
	--- a/lib/fast_html.ex
	+++ b/lib/fast_html.ex
	@@ -1,150 +1,151 @@
	# SPDX-FileCopyrightText: 2017-2019 myhtmlex authors <https://github.com/Overbryd/myhtmlex>
	+# SPDX-FileCopyrightText: 2019-2022 Pleroma Authors <https://pleroma.social>
	# SPDX-License-Identifier: LGPL-2.1-only

	defmodule :fast_html do
	@moduledoc """
	A module to decode html into a tree structure.
	"""

	@type tag() :: String.t() \| atom()
	@type attr() :: {String.t(), String.t()}
	@type attr_list() :: [] \| [attr()]
	@type comment_node() :: {:comment, String.t()}
	@type comment_node3() :: {:comment, [], String.t()}
	@type tree() ::
	{tag(), attr_list(), tree()}
	\| {tag(), attr_list(), nil}
	\| comment_node()
	\| comment_node3()
	@type format_flag() :: :html_atoms \| :nil_self_closing \| :comment_tuple3

	@doc """
	Returns a tree representation from the given html string.

	`opts` is a keyword list of options, the options available:
	* `timeout` - Call timeout. If pooling is used and the worker doesn't return
	the result in time, the worker will be killed with a warning.
	* `format` - Format flags for the tree.

	The following format flags are available:

	* `:html_atoms` uses atoms for known html tags (faster), binaries for everything else.
	* `:nil_self_closing` uses `nil` to designate void elements.
	For example `<br>` is then being represented like `{"br", [], nil}`.
	See http://w3c.github.io/html-reference/syntax.html#void-elements for a full list of void elements.
	* `:comment_tuple3` uses 3-tuple elements for comments, instead of the default 2-tuple element.

	## Examples
	iex> :fast_html.decode("<h1>Hello world</h1>")
	{:ok, [{"html", [], [{"head", [], []}, {"body", [], [{"h1", [], ["Hello world"]}]}]}]}

	iex> :fast_html.decode("Hello world", timeout: 0)
	{:error, :timeout}

	iex> :fast_html.decode("<span class='hello'>Hi there</span>")
	{:ok, [{"html", [],
	[{"head", [], []},
	{"body", [], [{"span", [{"class", "hello"}], ["Hi there"]}]}]}]}

	iex> :fast_html.decode("<body><!-- a comment --!></body>")
	{:ok, [{"html", [], [{"head", [], []}, {"body", [], [comment: " a comment "]}]}]}

	iex> :fast_html.decode("<br>")
	{:ok, [{"html", [], [{"head", [], []}, {"body", [], [{"br", [], []}]}]}]}

	iex> :fast_html.decode("<h1>Hello world</h1>", format: [:html_atoms])
	{:ok, [{:html, [], [{:head, [], []}, {:body, [], [{:h1, [], ["Hello world"]}]}]}]}

	iex> :fast_html.decode("<br>", format: [:nil_self_closing])
	{:ok, [{"html", [], [{"head", [], []}, {"body", [], [{"br", [], nil}]}]}]}

	iex> :fast_html.decode("<body><!-- a comment --!></body>", format: [:comment_tuple3])
	{:ok, [{"html", [], [{"head", [], []}, {"body", [], [{:comment, [], " a comment "}]}]}]}

	iex> html = "<body><!-- a comment --!><unknown /></body>"
	iex> :fast_html.decode(html, format: [:html_atoms, :nil_self_closing, :comment_tuple3])
	{:ok, [{:html, [],
	[{:head, [], []},
	{:body, [], [{:comment, [], " a comment "}, {"unknown", [], []}]}]}]}

	"""
	@spec decode(String.t(), format: [format_flag()]) ::
	{:ok, tree()} \| {:error, String.t() \| atom()}
	def decode(bin, opts \\ []) do
	flags = Keyword.get(opts, :format, [])
	timeout = Keyword.get(opts, :timeout, 10000)

	find_and_use_port({:decode, bin, flags}, timeout, opts)
	end

	@doc """
	Like `decode/2`, but for parsing [HTML fragments](https://html.spec.whatwg.org/multipage/parsing.html#parsing-html-fragments).

	`opts` is a keyword list of options, the options available are the same as in `decode/2` with addition of:
	* `context` - Name of the context element, defaults to `div`

	Example:
	iex> :fast_html.decode_fragment("rin is the <i>best</i> girl")
	{:ok, ["rin is the ", {"i", [], ["best"]}, " girl"]}
	iex> :fast_html.decode_fragment("rin is the <i>best</i> girl", context: "title")
	{:ok, ["rin is the <i>best</i> girl"]}
	iex> :fast_html.decode_fragment("rin is the <i>best</i> girl", context: "objective_truth")
	{:error, :unknown_context_tag}
	iex> :fast_html.decode_fragment("rin is the <i>best</i> girl", format: [:html_atoms])
	{:ok, ["rin is the ", {:i, [], ["best"]}, " girl"]}
	"""
	def decode_fragment(bin, opts \\ []) do
	flags = Keyword.get(opts, :format, [])
	timeout = Keyword.get(opts, :timeout, 10000)
	context = Keyword.get(opts, :context, "div")

	find_and_use_port({:decode_fragment, bin, flags, context}, timeout, opts)
	end

	@default_pool FastHtml.Pool
	defp find_and_use_port(term_command, timeout, opts) do
	command = :erlang.term_to_binary(term_command)

	pool =
	cond do
	pool = Keyword.get(opts, :pool) -> pool
	Application.get_env(:fast_html, :pool, enabled: true)[:enabled] -> @default_pool
	true -> nil
	end

	execute_command_fun = fn port ->
	send(port, {self(), {:command, command}})

	receive do
	{^port, {:data, res}} -> {:ok, res}
	after
	timeout ->
	{:error, :timeout}
	end
	end

	result =
	if pool do
	FastHtml.Pool.get_port(pool, execute_command_fun)
	else
	port = open_port()
	result = execute_command_fun.(port)
	Port.close(port)
	result
	end

	case result do
	{:ok, result} -> :erlang.binary_to_term(result)
	{:error, _} = e -> e
	end
	end

	def open_port do
	Port.open({:spawn_executable, Path.join([:code.priv_dir(:fast_html), "fasthtml_worker"])}, [
	:binary,
	{:packet, 4},
	:use_stdio,
	:exit_status
	])
	end
	end
	diff --git a/lib/fast_html/application.ex b/lib/fast_html/application.ex
	index 68b3c3b..0502701 100644
	--- a/lib/fast_html/application.ex
	+++ b/lib/fast_html/application.ex
	@@ -1,15 +1,18 @@
	+# SPDX-FileCopyrightText: 2019-2022 Pleroma Authors <https://pleroma.social>
	+# SPDX-License-Identifier: LGPL-2.1-only
	+
	defmodule FastHtml.Application do
	@moduledoc false

	use Application

	def start(_type, _args) do
	default_pool_config = Application.get_env(:fast_html, :pool, enabled: true)
	children = if default_pool_config[:enabled], do: [FastHtml.Pool], else: []

	Supervisor.start_link(children,
	strategy: :one_for_one,
	name: FastHtml.Supervisor
	)
	end
	end
	diff --git a/lib/fast_html/pool.ex b/lib/fast_html/pool.ex
	index 47a29e1..2032f28 100644
	--- a/lib/fast_html/pool.ex
	+++ b/lib/fast_html/pool.ex
	@@ -1,132 +1,135 @@
	+# SPDX-FileCopyrightText: 2019-2022 Pleroma Authors <https://pleroma.social>
	+# SPDX-License-Identifier: LGPL-2.1-only
	+
	defmodule FastHtml.Pool do
	@behaviour NimblePool
	@moduledoc """

	"""

	require Logger

	@doc false
	def child_spec(opts) do
	%{
	id: __MODULE__,
	start: {__MODULE__, :start_link, [opts]},
	type: :worker,
	restart: :permanent
	}
	end

	@doc """
	Starts the port pool.

	### Options
	- `:size` - Number of ports in the pool. Defaults to `System.schedulers_online/0` if not set.
	- `:name` - Registered name of the pool. Defaults to `#{__MODULE__}` if not set, set to `false` to not register the process.
	"""
	@type option :: {:size, pos_integer()} \| {:name, atom()}
	@spec start_link([option()]) :: term()
	def start_link(options) do
	{size, options} = Keyword.pop(options, :size, System.schedulers_online())
	NimblePool.start_link(worker: {__MODULE__, options}, pool_size: size)
	end

	@type pool :: atom() \| pid()
	@type result :: {:ok, term()} \| {:error, atom()}
	@spec get_port(pool(), (port() -> result())) :: result()
	def get_port(pool, fun) do
	NimblePool.checkout!(pool, :checkout, fn _from, port ->
	result = fun.(port)

	client_state =
	case result do
	{:ok, _} ->
	:ok

	{:error, reason} ->
	reason
	end

	send(port, {self(), {:connect, GenServer.whereis(pool)}})

	client_state =
	receive do
	{^port, :connected} -> client_state
	{:EXIT, ^port, reason} -> {:EXIT, reason}
	end

	{result, client_state}
	end)
	end

	@impl NimblePool
	@doc false
	def init_pool(state) do
	{name, options} =
	case Keyword.pop(state, :name) do
	{nil, state} -> {__MODULE__, state}
	{name, state} when is_atom(name) -> {name, state}
	{_, state} -> {nil, state}
	end

	if name, do: Process.register(self(), name)
	{:ok, options}
	end

	@impl NimblePool
	@doc false
	def init_worker(pool_state) do
	port = :fast_html.open_port()
	{:ok, port, pool_state}
	end

	@impl NimblePool
	@doc false
	def terminate_worker({:EXIT, reason}, port, pool_state) do
	Logger.warn(fn ->
	"[#{__MODULE__}]: Port #{port} unexpectedly exited with reason: #{reason}"
	end)

	{:ok, pool_state}
	end

	@impl NimblePool
	@doc false
	def terminate_worker(_reason, port, pool_state) do
	Port.close(port)
	{:ok, pool_state}
	end

	@impl NimblePool
	@doc false
	def handle_checkout(:checkout, {client_pid, _}, port, pool_state) do
	send(port, {self(), {:connect, client_pid}})

	receive do
	{^port, :connected} -> {:ok, port, port, pool_state}
	{:EXIT, ^port, reason} -> {:remove, {:EXIT, reason}}
	end
	end

	@impl NimblePool
	@doc false
	def handle_checkin(:timeout, _, _, pool_state), do: {:remove, :timeout, pool_state}

	@impl NimblePool
	@doc false
	def handle_checkin(_, _, port, pool_state), do: {:ok, port, pool_state}

	@impl NimblePool
	@doc false
	def handle_info({:EXIT, port, reason}, port), do: {:remove, {:EXIT, reason}}

	@impl NimblePool
	@doc false
	def handle_info({:EXIT, _, _}, port), do: {:ok, port}

	# Port sent data to the pool, this happens when the timeout was reached
	# and the port got disconnected from the client, but not yet killed by the pool.
	# Just discard the message.
	@impl NimblePool
	@doc false
	def handle_info({_sending_port, {:data, _}}, port), do: {:ok, port}
	end
	diff --git a/lib/mix/tasks/fast_html/bench.ex b/lib/mix/tasks/fast_html/bench.ex
	index 32823ee..1198615 100644
	--- a/lib/mix/tasks/fast_html/bench.ex
	+++ b/lib/mix/tasks/fast_html/bench.ex
	@@ -1,31 +1,34 @@
	+# SPDX-FileCopyrightText: 2019-2022 Pleroma Authors <https://pleroma.social>
	+# SPDX-License-Identifier: LGPL-2.1-only
	+
	if Mix.env() == :bench do
	defmodule Mix.Tasks.FastHtml.Bench do
	@moduledoc "Benchmarking task."

	use Mix.Task

	@input_dir "bench_fixtures"

	def run(_) do
	Application.ensure_all_started(:fast_html)

	inputs =
	Enum.reduce(File.ls!(@input_dir), %{}, fn input_name, acc ->
	input = File.read!(Path.join(@input_dir, input_name))
	Map.put(acc, input_name, input)
	end)

	Benchee.run(
	%{
	"fast_html" => fn input -> :fast_html.decode(input) end,
	"myhtmlex nif" => fn input -> Myhtmlex.Nif.decode(input) end,
	"html5ever nif" => fn input -> Html5ever.parse(input) end,
	"mochiweb_html" => fn input -> :mochiweb_html.parse(input) end
	},
	inputs: inputs,
	save: [path: "fast_html.bench"],
	load: "fast_html.bench"
	)
	end
	end
	end
	diff --git a/mix.exs b/mix.exs
	index a98ac53..06f1b7d 100644
	--- a/mix.exs
	+++ b/mix.exs
	@@ -1,116 +1,117 @@
	# SPDX-FileCopyrightText: 2017-2019 myhtmlex authors <https://github.com/Overbryd/myhtmlex>
	+# SPDX-FileCopyrightText: 2019-2022 Pleroma Authors <https://pleroma.social>
	# SPDX-License-Identifier: LGPL-2.1-only

	defmodule FastHtml.Mixfile do
	use Mix.Project

	def project do
	[
	app: :fast_html,
	version: "2.0.5",
	elixir: "~> 1.5",
	deps: deps(),
	package: package(),
	compilers: [:elixir_make] ++ Mix.compilers(),
	make_env: make_env(),
	make_error_message: make_error_message(),
	build_embedded: Mix.env() == :prod,
	start_permanent: Mix.env() == :prod,
	name: "FastHtml",
	description: """
	A module to decode HTML into a tree,
	porting all properties of the underlying
	library lexbor, being fast and correct
	in regards to the html spec.
	""",
	docs: docs()
	]
	end

	def package do
	[
	maintainers: ["Ariadne Conill", "rinpatch"],
	licenses: ["LGPL-2.1-only"],
	links: %{
	"GitLab" => "https://git.pleroma.social/pleroma/elixir-libraries/fast_html/",
	"Issues" => "https://git.pleroma.social/pleroma/elixir-libraries/fast_html/issues",
	"lexbor" => "https://github.com/lexbor/lexbor"
	},
	files: hex_files()
	]
	end

	def application do
	[
	extra_applications: [:logger],
	mod: {FastHtml.Application, []}
	]
	end

	defp deps do
	[
	# documentation helpers
	{:ex_doc, "~> 0.19", only: :dev},
	# benchmarking helpers
	{:benchee, "~> 1.0", only: :bench, optional: true},
	{:dialyxir, "~> 1.0", only: [:dev, :test], runtime: false},
	{:myhtmlex, "~> 0.2.0", only: :bench, runtime: false, optional: true},
	{:mochiweb, "~> 2.18", only: :bench, optional: true},
	{:html5ever,
	git: "https://github.com/rusterlium/html5ever_elixir.git", only: :bench, optional: true},
	{:nimble_pool, "~> 0.2.0"},
	{:elixir_make, "~> 0.4", runtime: false}
	]
	end

	defp docs do
	[
	main: "readme",
	extras: ["README.md", "CHANGELOG.md"]
	]
	end

	defp hex_files do
	# This is run every time mix is executed, so it will fail in the hex package,
	# therefore check if git is even available
	if File.exists?(".git") and System.find_executable("git") do
	{files, 0} = System.cmd("git", ["ls-files", "--recurse-submodules"])

	files
	\|> String.split("\n")
	# Last element is "", which makes hex include all files in the folder to the project
	\|> List.delete_at(-1)
	\|> Enum.reject(fn path ->
	Path.dirname(path) == "bench_fixtures" or
	(Path.dirname(path) != "priv" and String.starts_with?(Path.basename(path), "."))
	end)
	else
	[]
	end
	end

	defp otp_version do
	:erlang.system_info(:otp_release)
	\|> to_string()
	\|> String.to_integer()
	end

	defp otp_22_or_newer? do
	otp_version() >= 22
	end

	defp make_env do
	%{
	"OTP22_DEF" =>
	if otp_22_or_newer?() do
	"YES"
	else
	"NO"
	end
	}
	end

	defp make_error_message,
	do:
	"Please check you have: a C compiler, GNU\Make, CMake and Erlang development headers installed before reporting an issue."
	end
	diff --git a/mix.lock.license b/mix.lock.license
	index cd7ec38..e7038e9 100644
	--- a/mix.lock.license
	+++ b/mix.lock.license
	@@ -1,4 +1,4 @@
	SPDX-FileCopyrightText: 2017-2019 myhtmlex authors <https://github.com/Overbryd/myhtmlex>
	-SPDX-FileCopyrightText: 2019-2022 Pleroma Authors
	+SPDX-FileCopyrightText: 2019-2022 Pleroma Authors <https://pleroma.social>

	SPDX-License-Identifier: LGPL-2.1-only
	diff --git a/test/fast_html_test.exs b/test/fast_html_test.exs
	index 588732d..d243c0a 100644
	--- a/test/fast_html_test.exs
	+++ b/test/fast_html_test.exs
	@@ -1,168 +1,169 @@
	# SPDX-FileCopyrightText: 2017-2019 myhtmlex authors <https://github.com/Overbryd/myhtmlex>
	+# SPDX-FileCopyrightText: 2019-2022 Pleroma Authors <https://pleroma.social>
	# SPDX-License-Identifier: LGPL-2.1-only

	defmodule :fast_html_test do
	use ExUnit.Case
	doctest :fast_html

	test "doesn't segfault when <!----> is encountered" do
	assert {:ok, [{"html", _attrs, _children}]} = :fast_html.decode("<div> <!----> </div>")
	end

	test "builds a tree, formatted like mochiweb by default" do
	assert {:ok,
	[
	{"html", [],
	[
	{"head", [], []},
	{"body", [],
	[
	{"br", [], []}
	]}
	]}
	]} = :fast_html.decode("<br>")
	end

	test "builds a tree, html tags as atoms" do
	assert {:ok,
	[
	{:html, [],
	[
	{:head, [], []},
	{:body, [],
	[
	{:br, [], []}
	]}
	]}
	]} = :fast_html.decode("<br>", format: [:html_atoms])
	end

	test "builds a tree, nil self closing" do
	assert {:ok,
	[
	{"html", [],
	[
	{"head", [], []},
	{"body", [],
	[
	{"br", [], nil},
	{"esi:include", [], []}
	]}
	]}
	]} = :fast_html.decode("<br><esi:include />", format: [:nil_self_closing])
	end

	test "builds a tree, multiple format options" do
	assert {:ok,
	[
	{:html, [],
	[
	{:head, [], []},
	{:body, [],
	[
	{:br, [], nil}
	]}
	]}
	]} = :fast_html.decode("<br>", format: [:html_atoms, :nil_self_closing])
	end

	test "attributes" do
	assert {:ok,
	[
	{:html, [],
	[
	{:head, [], []},
	{:body, [],
	[
	{:span, [{"id", "test"}, {"class", "foo garble"}], []}
	]}
	]}
	]} =
	:fast_html.decode(~s'<span id="test" class="foo garble"></span>',
	format: [:html_atoms]
	)
	end

	test "single attributes" do
	assert {:ok,
	[
	{:html, [],
	[
	{:head, [], []},
	{:body, [],
	[
	{:button, [{"disabled", ""}, {"class", "foo garble"}], []}
	]}
	]}
	]} =
	:fast_html.decode(~s'<button disabled class="foo garble"></span>',
	format: [:html_atoms]
	)
	end

	test "text nodes" do
	assert {:ok,
	[
	{:html, [],
	[
	{:head, [], []},
	{:body, [],
	[
	"text node"
	]}
	]}
	]} = :fast_html.decode(~s'<body>text node</body>', format: [:html_atoms])
	end

	test "broken input" do
	assert {:ok,
	[
	{:html, [],
	[
	{:head, [], []},
	{:body, [],
	[
	{:a, [{"<", ""}], [" asdf"]}
	]}
	]}
	]} = :fast_html.decode(~s'<a <> asdf', format: [:html_atoms])
	end

	test "custom namespaced tags" do
	assert {:ok,
	[
	{:html, [],
	[
	{:head, [], []},
	{:body, [],
	[
	{"esi:include", [], []}
	]}
	]}
	]} = :fast_html.decode(~s'<esi:include />', format: [:html_atoms, :nil_self_closing])
	end

	test "html comments" do
	assert {:ok,
	[
	{:html, [],
	[
	{:head, [], []},
	{:body, [],
	[
	comment: " a comment "
	]}
	]}
	]} = :fast_html.decode(~s'<body><!-- a comment --></body>', format: [:html_atoms])
	end

	test "doesn't go into an infinite loop when there are more than one root tags" do
	assert {:ok, [{:comment, " a comment "}, {"html", [], [{"head", [], []}, {"body", [], []}]}]} ==
	:fast_html.decode("<!-- a comment --> <html> </html>")
	end

	test "doesn't return attribute name in attribute value when the latter is empty" do
	assert :fast_html.decode_fragment("<meta content=\"\"/>") ==
	{:ok, [{"meta", [{"content", ""}], []}]}
	end
	end
	diff --git a/test/test_helper.exs b/test/test_helper.exs
	index 2f44730..021039e 100644
	--- a/test/test_helper.exs
	+++ b/test/test_helper.exs
	@@ -1,4 +1,5 @@
	# SPDX-FileCopyrightText: 2017-2019 myhtmlex authors <https://github.com/Overbryd/myhtmlex>
	+# SPDX-FileCopyrightText: 2019-2022 Pleroma Authors <https://pleroma.social>
	# SPDX-License-Identifier: LGPL-2.1-only

	ExUnit.start()

File Metadata

Mime Type: text/x-diff
Expires: Sun, Nov 24, 10:33 PM (1 d, 7 h)
Storage Engine: blob
Storage Format: Raw Data
Storage Handle: 39491
Default Alt Text: (45 KB)

No OneTemporaryActions

View Options

File Metadata

Event Timeline

No OneTemporary
Actions